Quantification of Active Sites for the Determination of Methanol

However, the difficulties involved in determining active surface site densities on ... number of active metal oxide surface sites available for methan...
0 downloads 0 Views 127KB Size
Download PDF
Langmuir 2001, 17, 6175-6184

6175

Quantification of Active Sites for the Determination of Methanol Oxidation Turn-over Frequencies Using Methanol Chemisorption and in Situ Infrared Techniques. 2. Bulk Metal Oxide Catalysts Loyd J. Burcham,† Laura E. Briand,‡ and Israel E. Wachs*,§ Zettlemoyer Center for Surface Studies and Department of Chemical Engineering, Lehigh University, Bethlehem, Pennsylvania 18015, and Centro de Investigacion y Desarrollo en Procesos Cataliticos, Universidad Nacional de La Plata, Calle 47 No. 257, (1900) La Plata, Argentina Received January 4, 2001. In Final Form: May 8, 2001 Bulk metal oxide catalysts, especially bulk mixed-metal molybdates such as Fe2(MoO4)3, often exhibit high methanol oxidation activity and selectivity. However, the difficulties involved in determining active surface site densities on these catalysts have, in the past, generally prevented side-by-side comparisons of their intrinsic activities, or turn-over frequencies (TOFs). In the present study, high temperature (110 °C) methanol chemisorption and in-situ infrared spectroscopy have been employed to directly and quantitatively determine the number of active metal oxide surface sites available for methanol oxidation. The IR spectra indicate that methanol chemisorption on these catalysts produces both associatively adsorbed, intact Lewis-bound surface methanol species (CH3OHads, species I) on acidic sites, as well as dissociatively adsorbed surface methoxy species (-OCH3, species II) on less acidic or basic sites. In fact, the Lewis acidity of bulk mixed-metal molybdates relative to the methanol probe molecule was found to decrease as follows: Fe2(MoO4)3, NiMoO4 (species I predominates) > MnMoO4, CoMoO4, ZnMoO4, Al2(MoO4)3 > Ce(MoO4)2 > Bi2Mo3O12 > Zr(MoO4)2 (species II predominates). It also appears that Mo cations are the primary methanol chemisorption sites in many of the bulk mixed-metal molybdates, including commercially important Fe2(MoO4)3. By quantifying the surface concentrations of the adsorbed methoxylated reaction intermediates from the IR spectra, it was then possible to normalize the catalytic methanol oxidation activities for the calculation of TOFs. The methanol oxidation TOFs of bulk molybdates were shown to be relatively similar to those of model supported catalysts with the same co-cation (e.g., MoO3/NiO vs NiMoO4)spossibly due to the formation of a “monolayer” of surface molybdenum oxide species on the surfaces of the bulk metal molybdates. In addition, the bulk mixed-metal molybdates were found to exhibit the same ligand effect as that discovered previously in supported metal oxide catalysts, in which the TOF generally decreases with increasing ligand cation electronegativity due to electronic variations in localized M-O-Ligand bonds.

Introduction Bulk metal oxide catalysts are commercially important for the production of formaldehyde, especially Fe2(MoO4)3MoO3 mixtures and other bulk mixed-metal molybdates.1,2 However, the development of fundamental structurereactivity relationships in these systems has been limited by the difficulty involved in determining the number of active metal oxide surface sites present during reaction. Such knowledge is critical for the calculation of methanol oxidation turn-over frequencies over bulk metal oxides (TOF ) methanol molecules converted to formaldehyde per second per active surface metal oxide site). The TOFs themselves are important because they provide the best basis for comparison of intrinsic catalytic activities between different catalysts.2-4 Active surface site density * Corresponding author. Phone: 610-758-4274. Fax: 610-7586555. E-mail: [email protected]. † Present address: Congoleum Corporation, P.O. Box 3127, Mercerville, NJ 08619. ‡ Centro de Investigacion y Desarrollo en Procesos Cataliticos. § Zettlemoyer Center for Surface Studies and Department of Chemical Engineering. (1) Gerberich, H. R.; Seaman, G. C. Formaldehyde. In Kirk-Othmer Encyclopedia of Chemical Technology, 4th ed.; John Wiley and Sons: New York, 1994; Vol. 11, pp 929-951. (2) Tatiboue¨t, J. M. Appl. Catal. A 1997, 148, 213. (3) Thomas, J. M.; Thomas, W. J. Principles and Practice of Heterogeneous Catalysis; VCH Publishers: New York, 1997.

measurements for metal and supported metal catalysts are often made using the chemisorption of CO, H2, and O2 probe molecules to quantify the active metal surface area,2-4 but the chemisorption of these traditional probe molecules on metal oxides is more difficult. Instead, the use of methanol as a chemisorption probe molecule provides a more feasible method for quantifying active surface site densities in metal oxide catalysts, since methanol is easily chemisorbed onto metal oxide surfaces. Moreover, the chemisorbed methoxylated surface species have been shown to be the reactive surface intermediates in methanol oxidation.5,6 Thus, methanol is also the ideal probe molecule for titration of the specific metal oxide surface sites involved in methanol oxidation. In part 1 of the present two-paper series,7 an IR spectroscopy-based methodology was developed for quantitative methanol chemisorption using supported metal oxides as model systems. Supported metal oxide catalysts consist of an active metal oxide (MoO3, V2O5, Cr2O3, Nb2O5, Re2O7, WO3, etc.) molecularly dispersed as a spectroscopically distinguishable, two-dimensional surface metal oxide overlayer on a high surface area oxide support (Al2O3, (4) Ribeiro, F. H.; Schach von Wittenau, A. E.; Bartholomew, C. H.; Somorjai, G. A. Catal. Rev.-Sci. Eng. 1997, 39, 49. (5) Holstein, W. L.; Machiels, C. J. J. Catal. 1996, 162, 118. (6) Burcham, L. J.; Wachs, I. E. Catal. Today 1999, 49, 467. (7) Burcham, L. J.; Briand, L. E.; Wachs, I. E. Langmuir 2001, 17, 6164.

10.1021/la010010t CCC: $20.00 © 2001 American Chemical Society Published on Web 09/08/2001

6176

Langmuir, Vol. 17, No. 20, 2001

SiO2, TiO2, ZrO2, CeO2, MgO, etc.). For such supported metal oxide catalysts, previous studies had taken the number of metal oxide metal atoms deposited on the oxide support surface to be the number of active surface sites.8,9 However, using methanol chemisorption and IR techniques it was found in part 17 that a steric limitation of about 0.3 methoxylated surface species exists per active deposited metal oxide metal atom across all supported metal oxides. Methanol oxidation TOFs calculated using these new methanol chemisorption surface site densities were about 3 times higher than the TOFs previously estimated using the total number of deposited metal oxide metal atoms. Nevertheless, the fundamental support effect observed previously (TOFs for MoO3 and V2O5 supported on oxides of Zr ∼ Ce > Ti > Al . Si) remained virtually unchanged as a general trend and correlated with the support cation electronegativitysmost likely through the bridging M-O-Support bonds.7 For bulk metal oxides, it is generally more difficult to distinguish, spectroscopically or otherwise, the active surface metal oxide sites from the metal-oxygen bonds present within the framework lattice structure. Bulk metal oxide active site densities and methanol oxidation TOF's have been occasionally reported using crystal structure models10 or by gravimetric methanol chemisorption,11-13 but a generally accepted method has not yet been developed. Many authors14-23 propose oxygen chemisorption as a general method, in which the bulk (or supported) catalyst is reduced only at the surface and then reoxidized to determine the number of surface active metal oxide sites by the amount of oxygen consumed. However, this technique is very sensitive to over-reduction beyond the surface layer and is rather indirect.17,24-26 Another proposed method is the benzaldehyde-ammonia titration (BAT), in which benzaldehyde is selectively chemisorbed only on basic sites associated with inactive oxide supports (oxides of Al, Ti, Zr, etc.) and the benzonitrile produced by subsequent NH3 flow is used to quantify the exposed inactive and active surface areas.27-29 However, this (8) Wachs, I. E.; Deo, G.; Vuurman, M. A.; Hu, H.; Kim, D. S.; Jehng, J. M. J. Mol. Catal. 1993, 82, 443. (9) Wachs, I. E. Catal. Today 1996, 27, 437. (10) Hu, H.; Wachs, I. E. J. Phys. Chem. 1995, 99, 10911. (11) Farneth, W. E.; McCarron, E. M.; Sleight, A. W.; Staley, R. H. Langmuir 1987, 3, 217. (12) Farneth, W. E.; Staley, R. H.; Sleight, A. W. J. Am. Chem. Soc. 1986, 108, 2327. (13) Farneth, W. E.; Ohuchi, F.; Staley, R. H.; Chowdhry, U.; Sleight, A. W. J. Phys. Chem. 1985, 89, 2493. (14) Desikan, A. N.; Huang, L.; Oyama, S. T. J. Phys. Chem. 1991, 95, 10050. (15) Chary, K. V. R.; Vijayakumar, V.; Rao, P. K. Langmuir 1990, 6, 1549. (16) Reddy, B. M.; Chary, K. V. R.; Rama Rao, B.; Subrahmanyam, V. S.; Sunandana, C. S.; Nag, N. K. Polyhedron 1986, 5, 191. (17) Majunke, F.; Baerns, M.; Baiker, A.; Koeppel, R. A. Catal. Today 1994, 20, 53. (18) Rao, P. K.; Narasimha, K. ACS Symp. Ser. 1993, 523, 231. (19) Chary, K. V. R. J. Chem. Soc., Chem. Commun. 1989, 104. (20) Reddy, B. M.; Manohar, B.; Reddy, E. P. Langmuir 1993, 9, 1781. (21) Chary, K. V. R.; Rao, B. R.; Subrahmanyam, V. S. Appl. Catal. 1991, 74, 1. (22) Arena, F.; Frusteri, F.; Parmaliana, A. Appl. Catal. A 1999, 176, 189. (23) Nag, N.; Chary, K.; Subrahmanyam, V. J. Chem. Soc., Chem. Commun. 1986, 1147. (24) Faraldos, M.; Anderson, J. A.; Banares, M. A.; Fierro, J. L. G.; Weller, S. W. J. Catal. 1997, 168, 110. (25) Tanabe, S.; Davis, H. E.; Wei, D.; Weber, R. S. In Studies in Surface Science and Catalysis (Proceedings of the 11th International Congress on Catalysis); Hightower, J. W., Delgass, W. N., Iglesia, E., Bell, A. T., Eds.; Elsevier: Amsterdam, 1996; Vol. 101, p 337. (26) Deo, G.; Wachs, I. E.; Haber, J. Crit. Rev. Surf. Chem. 1994, 4, 141.

Burcham et al.

indirect method also suffers from nonselective adsorption and assumptions regarding the spatial dimensions of the surface active sites in the active surface area.26 In all of these cases, the probe molecules employed are quite different from the organic molecules typically involved in oxidation reactions. The works of Farneth et al.11-13 on bulk MoO3 and Sleight et al.30 on Fe2(MoO4)3 were perhaps the first to suggest that quantitative methanol chemisorption may be a viable method for determining the active site density in bulk metal oxides. Gas-phase methanol is well-known to chemically adsorb on oxide surfaces at low temperatures ( MnMoO4, CoMoO4, ZnMoO4, Al2(MoO4)3 > Ce(MoO4)2 > Bi2Mo3O12 > Zr(MoO4)2 (species II predominates). In general, these bulk mixed-metal molybdates, in similarity to the supported molybdena catalysts studied in part 1,7 have a strong tendency to adsorb methanol as Lewis-bound adducts rather than as the dissociated surface methoxy species favored by more basic oxide surfaces. Moreover, the Lewis basicity of the Fe, Ni, Bi, Zn, and Zr cocations relative to the methanol probe molecule was confirmed by the presence of ionic methoxides on the corresponding pure oxides (lowest C-H stretching bands at ∼2800-2820 cm-1, see Table 1). As previously noted, the absence of these methoxide signals in the spectra of the mixed-metal molybdates suggests that the basic cocations are relatively inaccessible for methanol chemisorption and that Mo cations are the primary adsorption centers for methanol. 4.2. Methanol Chemisorption for Quantification of Active Metal Oxide Surface Sites. The high tem(53) (a) Turek, A. M.; Wachs, I. E.; DeCanio, E. J. Phys. Chem. 1992, 96, 5000. (b) Datka, J.; Turek, A. M.; Jehng, J.-M.; Wachs, I. E. J. Catal. 1992, 135, 186.

Langmuir, Vol. 17, No. 20, 2001 6181

Figure 3. Direct, side-by-side comparisons of the methanol oxidation TOFs in bulk and supported molybdena catalysts using TOFs based on the methanol chemisorption site densities.

perature (110 °C) methanol chemisorption and in-situ IR technique used in the present study overcomes the uncertainties associated with the more complicated site density measurement procedures discussed in the Introduction by directly measuring the number of active surface intermediates in bulk metal oxides. Interestingly, Lewisbound water (IR band at ∼1610 cm-1) was found to be easily displaced by the adsorbed methoxylated surface species in part 1, independently of whether the surface water was initially present adventitiously or was formed by condensation of surface hydroxyls following dissociative methanol chemisorption.7 Quantitatively, such an evolution of water from the oxide surface may allow for the assumption of complete water loss in high temperature (110 °C) gravimetric methanol adsorption studies. However, at lower adsorption temperatures the possibility of forming nondesorbing, surface-bound water from dissociative methanol chemisorption requires that the methoxylated surface species be quantified directly by, for example, IR spectroscopy (CeO2,47 ZnO,MgO,44 MoO3,41 SiO2,38 and ZrO248), or that the water desorbed upon adsorption be independently measured if gravimetric methods are used.11-13,50 The technique used in the present study not only determines the number of adsorbed methoxylated surface intermediates directly by IR spectroscopy, but it also minimizes the amount of physisorbed methanol and physisorbed water by performing the methanol chemisorption at elevated temperature and under vacuum. The active site densities and methanol oxidation TOFs presented in Table 3 for bulk mixed-metal molybdates and for pure bulk metal oxides provide unique new insights into the intrinsic catalysis of bulk metal oxides. For instance, the intrinsic activities (TOFs) of commercially relevant bulk molybdates are compared graphically in Figure 3 with each other and with the TOFs of the corresponding supported molybdena catalysts (data for the supported molybdena catalysts are taken from part 1 of the present two-paper series7). For the first time, it is possible to see that the intrinsic TOFs of model supported metal oxide catalysts are relatively similar to those of bulk mixed molybdates with the same co-cation (e.g., MoO3/NiO vs NiMoO4). There certainly are not orders of magnitude differences between such catalyst pairs, although large differences are observed between pairs (e.g., Mo-Al vs Mo-Zr) and among the bulk molybdates. Furthermore, such a fundamental observation raises the possibility that the same ligand effect that is respon-

6182

Langmuir, Vol. 17, No. 20, 2001

Figure 4. Methanol chemisorption site density vs ligand cation electronegativity in bulk and supported molybdena catalysts.

sible for the orders of magnitude variation of TOF in supported metal oxide catalysts (see part 17) may also apply to bulk mixed molybdates. More specifically, and by analogy to supported metal oxides, the electronegativity of the ligand (or support) cation likely controls the TOF in both bulk and supported molybdena catalysts, where more electropositive ligand cations correspond to greater TOFs. The specific molecular structures and coordinations of the molybdenum and cocation sites appear to be secondary to this localized electronic ligand effect imparted by the co-cation toward the surface active Mo cation. Some tests of this hypothesis are given in Figures 4-6, the establishment of important noncorrelations being the subject of the first two of these figures. For instance, Figure 4 shows that the methanol chemisorption surface site densities, which average ∼3.9 µmol of methoxylated surface intermediates per m2, are independent of the ligand cation electronegativity in both supported and bulk molybdena catalysts. Only a mild increase in surface site density is observed for the bulk molybdates relative to the supported molybdena catalysts. Note that Ns for MoO3/ SiO2 includes inactive spectator methoxy surface sites on exposed silica cations, and Ns for Bi2Mo3O12 may not be very accurate due to its extremely low surface area of 0.22 m2/g. The variation in Ns for these samples, which is about ( 3.2 µmols/m2, also means that the average value of Ns (3.9 µmol/m2) should only be taken as a rough average of methanol chemisorption site densities on oxides. Such an average figure has merit in indicating the relative scale of Ns, which is much lower than the surface sites densities reported for metals. However, calculations based on the average Ns value for methanol adsorbed on oxides should only be made in conjunction with its deviation, where minimum and maximum values of Ns (see Table 3) are used to “bound” the calculation. Additionally, while the error in Ns is most likely greater for samples with low surface area (e.g., Bi2Mo3O12), the TOFs are not influenced directly by surface area measurements because the activity and Ns values are both normalized to the same surface area value, which is canceled upon calculation of TOF. At any rate, Figure 4 indicates that the specific ligand cation is not significantly related to site density. Figure 5 reveals that the methanol oxidation TOF in bulk and supported molybdena catalysts is also not related to surface site density. This is expected based on the accepted mechanism for the unimolecular, single site decomposition of the adsorbed surface methoxy intermediates to formaldehyde,2,5,54 and reactions requiring more than one metal oxide site would have generated higher TOFs for catalysts with higher Ns values. Such a finding

Burcham et al.

Figure 5. Methanol oxidation TOF vs methanol chemisorption site density in bulk and supported molybdena catalysts.

Figure 6. The ligand effect in bulk and supported molybdena catalysts.

further suggests that methanol chemisorption is a more reliable method for counting active sites than the BAT method, which has been applied to supported molybdena catalysts by Niwa et al.27-29,55 Using the BAT technique, these authors found that the TOF values indicated the need for two metal oxide sites (linear increase in TOF with MoO3 loading)sin contradiction with the abovementioned and fairly well-verified single site mechanism.2,5,54 The clearest evidence that an electronic ligand effect controls the methanol oxidation TOF is given in Figure 6, where it can be seen that the TOF generally decreases with increasing ligand cation electronegativity. Note that the fitted exponential curve is added mainly to indicate the trend and does not imply a strictly exponential relationship, especially considering the concentration of data points in the region of higher ligand electronegativity. In particular, the correlation is rather scattered for ligand cation electronegativities below a value of about 1.6, and there is very little trend between TOF and ligand cation electronegativity above this value. Overall, however, Figure 6 still shows that low ligand cation electronegativities generally correspond to higher methanol oxidation TOFs. Moreover, this ligand effect is exhibited by both supported and bulk molybdate catalysts, the TOFs in the former having been extrapolated to 380 °C from their values at 250 °C (Arrhenius extrapolation based on data (54) Wachs, I. E. In Catalysis; Spivey, J. J., Ed.; The Royal Society of Chemistry: Cambridge, 1997; Vol. 13m pp 37-54. (55) Matsuoka, Y.; Niwa, M.; Murakami, Y. J. Phys. Chem. 1990, 94, 1477.

CH3OH Chemisorption on Metal Oxide Catalysts

given in part 17) in order to allow for direct comparison with the lower surface area bulk molybdates. The origin of the ligand effect has been investigated in supported metal oxides by Wachs et al.6,8-10,36,54,56,57 and Bell et al.,58 and a detailed discussion is given in part 1 of the present two-paper series.7 Well-characterized insitu molecular structural studies have shown that structural variations in the number of M ) O double bonds, the number of M-O-M polymerized bonds, and the coordination of the active metal oxide metal atom have far less impact on the TOF than does the choice of support oxide.6,8-10,36,54,56,57 In addition, it appears that the bridging M-O-Support bond (M ) Mo, V, etc.) is the dominant reactive center in methoxy hydrogen abstraction during the formation of formaldehyde, based on recent quantum chemical calculations.59,60 Thus, the fundamental origin of the ligand effect in supported metal oxides appears to be related to the electronegativity of the support cation and its effect upon the redox behavior of the M-OSupport bond. Note that Sanderson electronegativity is preferred for its better handling of transition metal cations, and tabulated values are available.61 For the bulk metal molybdates, no such detailed surface models exist for describing the molecular structures of the active surface sites. In fact, studies of supported metal oxide catalysts have often been motivated by their applicability as “model” catalysts for these less tractable bulk oxides, and structure-reactivity relationships derived from the more easily characterized supported systems were expected to be extended by analogy to bulk oxides.2,8 Unfortunately, relatively little progress has been made in connecting the studies on model supported metal oxide catalysts to their bulk analogues. In the present study, the methanol chemisorption surface site density and corresponding methanol oxidation TOFs have been successfully and directly compared for the bulk and supported molybdate catalytic systems. The similarity in TOF between supported and bulk molybdate pairs (e.g., MoO3/Al2O3 vs Al2(MoO4)3, etc.) and the correlation of the TOF to ligand cation electronegativity in both supported and bulk systems strongly suggests that the ligand effect originates from the same source in each system. In fact, there is substantial evidence (XRD62 and reaction-induced metal oxide spreading experiments observed with in-situ Raman spectroscopy63) suggesting that a “monolayer” of surface molybdenum oxide species exists on the surfaces of the bulk metal molybdates due to the lower surface energy of ModO bonds relative to surface hydroxyls.63 Therefore, by analogy to the structure-reactivity relationships derived for supported metal oxide catalysts, it may be concluded that electronic variations in localized M-O-Ligand bonds are mainly responsible for differences in TOF in bulk metal oxides such as molybdates. The methanol chemisorption and infrared technique used in the present study also has critical implications for the kinetic modeling of the methanol oxidation reaction, (56) Wachs, I. E.; Weckhuysen, B. M. Appl. Catal. A 1997, 157, 67. (57) Wachs, I. E.; Deo, G.; Juskelis, M. V.; Weckhuysen, B. M. In Dynamics of Surfaces and Reaction Kinetics in Heterogeneous Catalysis; Froment, G. F., Waugh, K. C., Eds.; Elsevier: Amsterdam, 1997; pp 305-314. (58) Khodakov, A.; Olthof, B.; Bell, A. T.; Iglesia, E. J. Catal. 1999, 181, 205. (59) Weber, R. S. J. Phys. Chem. 1994, 98, 2999. (60) Tran, K.; Hanning-Lee, M. A.; Biswas, A.; Stiegman, A. E.; Scott, G. W. J. Am. Chem. Soc. 1995, 117, 2618. (61) (a) Sanderson, R. T. J. Chem. Educ. 1988, 65, 112. (b) Sanderson, R. T. Inorg. Chem. 1986, 25, 3518. (62) Fagherazzi, G.; Pernicone, N. J. Catal. 1970, 16, 321. (63) Wang, C.-B.; Cai, Y.; Wachs, I. E. Langmuir 1999, 15, 1223.

Langmuir, Vol. 17, No. 20, 2001 6183

in addition to its importance in developing the abovediscussed fundamental insights regarding bulk metal oxide Lewis-acidity correlations and bulk metal oxide methanol oxidation TOF correlations. Specifically, recent in-situ methanol oxidation studies over oxide catalysts64 showed that the fractional surface coverage of methoxylated surface intermediates relative to saturation coverage is between 0.3 and 0.68 under steady-state reaction conditions when chemisorption site densities are used to define the saturation coverage. Previous authors often assumed that the reaction proceeds in the low surface coverage regime (fractional coverage less than 0.2) when constructing kinetic models.5 Therefore, the use of methanol chemisorption site densities becomes very important for kinetic modeling, where it can no longer be assumed that methanol oxidation over oxides proceeds exclusively within the low surface coverage regime. Last, note that it was shown in part 17 that methanol oxidation TOFs are not especially dependent upon the ratio of species I to species II. This conclusion was based upon the fact that very different TOFs were found for supported metal oxide catalysts possessing similar distributions of methoxylated surface species I and species II,7 and that easy interconversion between various types of methoxylated surface species has been documented in other metal oxides.48 Therefore, use of the total number of methoxylated surface species (species I plus species II) for calculating the active surface site densities and TOFs in metal oxides is a reasonable quantification procedure and has been utilized throughout the present investigation. This procedure is certainly preferred over the much less reliable deconvolution and curve-fitting techniques required for separate quantifications of species I and species II (see examples in part 17). Nevertheless, it is worth qualitatively noting that the different ratios of methoxylated surface species may be related to formaldehyde selectivity. For example, low selectivity is generally obtained at high conversions in vanadia-based catalysts due to the readsorption of formaldehyde product on vanadia sites and their further oxidation to carbon oxides, whereas molybdena-based catalysts generally maintain their selectivity at high conversions.1,2 Therefore, it is possible that the presence of intact, strongly Lewiscoordinated methanol species (species I) prevents or blocks the readsorption of formaldehyde product in molybdena catalysts at high conversions. 5. Conclusions Bulk metal oxide and bulk mixed-metal molybdate catalysts, many of which exhibit high methanol oxidation activity and selectivity, have been studied in the present investigation by high temperature (110 °C) methanol chemisorption and in-situ infrared spectroscopy. This technique was found to serve as both a probe of the surface chemistry of bulk metal oxide catalysts and as a powerful method for determining the surface density of active metal oxide sites available for methanol oxidation. Unlike more complicated site density measurement procedures such as oxygen chemisorption or the BAT method, this technique determines the number of adsorbed methoxylated surface intermediates directly by IR spectroscopy. Additionally, the amount of physisorbed methanol and physisorbed water is minimized by performing the methanol chemisorption at elevated temperature and under vacuum. (64) (a) Burcham, L. J. Ph.D. Dissertation; Lehigh University: Bethlehem, PA, 2000; Chapter 5. (b) Burcham, L. J.; Badlani, M.; Wachs, I. E. J. Catal. 2000. Submitted for publication.

6184

Langmuir, Vol. 17, No. 20, 2001

As a probe of surface chemistry, the IR spectra indicate that methanol chemisorption on bulk mixed-metal oxide catalysts produces both associatively adsorbed, intact Lewis-bound surface methanol species (CH3OHads, species I) on acidic sites, as well as dissociatively adsorbed surface methoxy species (-OCH3, species II) on less acidic or basic sites. Differences in the C-H stretching band frequencies of these adsorbed methoxylated species were observed between basic oxides (Bi2O3, CeO2, Fe2O3, NiO, ZnO, and ZrO2) and acidic oxides (Al2O3, MoO3, Nb2O5, SiO2, TiO2, WO3, and V2O5) and permit identification of the specific metal cation involved in methanol adsorption on mixed oxides. In fact, it appears that Mo cations are the adsorption sites in many of the bulk mixed-metal molybdates, including the commercially important Fe2(MoO4)3. These Mo sites also appear to have a strong tendency to adsorb methanol as Lewis-bound species I, in similarity with the supported molybdena catalysts studied in part 1.7 More generally, the Lewis acidity of bulk mixed-metal molybdates relative to the methanol probe molecule was found to decrease as follows: Fe2(MoO4)3, NiMoO4 (species I predominates) > MnMoO4, CoMoO4, ZnMoO4, Al2(MoO4)3 > Ce(MoO4)2 > Bi2Mo3O12 > Zr(MoO4)2 (species II predominates). By spectroscopically quantifying the amount of these adsorbed methoxylated surface intermediates (species I and species II, collectively) present on the bulk metal oxides after saturation of the surfaces with methanol, it was possible to determine the number of active metal oxide surface sites available for methanol oxidation. Then, using methanol oxidation TOFs determined from these methanol chemisorption site densities it became possible to compare the intrinsic catalytic activities of different bulk metal oxides with themselves and, perhaps for the first time, with supported metal oxides (using data from part 17). In particular, the intrinsic TOFs of bulk molybdates were

Burcham et al.

shown to be relatively similar to those of model supported catalysts with the same cocation (e.g., MoO3/NiO vs NiMoO4), possibly due to the formation of a “monolayer” of surface molybdenum oxide species on the surfaces of the bulk metal molybdates. Moreover, the same electronic ligand effect appears to control the methanol oxidation TOF in both bulk as well as supported metal oxide catalysts, since TOF generally decreases with increasing ligand cation electronegativity in both groups. Therefore, by analogy to the structure-reactivity relationships derived for supported metal oxide catalysts it may be concluded that electronic variations in localized M-OLigand bonds are mainly responsible for differences in TOF in bulk metal oxides such as molybdates. Variations in the ratios of methoxylated surface species I to species II further suggest that the presence of intact, strongly Lewis-coordinated methanol species (species I) may prevent or block the readsorption of formaldehyde product in molybdena catalysts at high conversions to account for the favorable selectivity behavior of these systems. Last, the methanol chemisorption surface site densities themselves may influence future kinetic modeling of the methanol oxidation reaction, since recent studies64 indicate that a much larger fractional surface coverage of adsorbed methoxylated surface reaction intermediates are present during methanol oxidation than is generally assumed.5 As a result, the methanol chemisorption/infrared technique presented in this study may have additional applications beyond its present use for calculating methanol oxidation TOFs. Acknowledgment. The authors gratefully acknowledge the United States Department of Energy, Basic Energy Sciences (Grant DEFG02-93ER14350), for financial support of this work. LA010010T